1. The pilot conducted a low-level return transit to the replenishment point that was outside the pre-planned safe transit route.
2. The pilot did not see the powerline, or did not see the powerline in time to avoid a collision.

The damage to the right door frame corroborated the nature and distribution of the damaged perspex windscreen fragments along the wreckage trail and indicated that the helicopter struck the powerlines. The wirestrike was at a position consistent with that at which the property manager estimated he observed ‘glitter’ emanate from the helicopter.

It was likely that following the initial contact with the powerlines, the cockpit windscreen deformed sufficiently to capture the powerlines around the cockpit door hinge area. That would have prevented the powerlines from sliding up the windscreen towards the mast, or down towards the landing gear skids, before severing, and would explain the lack of any powerline damage or markings to the remainder of the helicopter.

The investigation considered whether a helicopter or associated system fault might have been implicated in the circumstances leading to the wirestrike. In that regard, there was no evidence of any technical or other failure of the helicopter or its associated systems prior to the ground impact, and the engine performed normally under test throughout the normal power range. That and the quality and quantity of the fuel remaining onboard the helicopter indicated that it was unlikely that any airframe, engine or system fault had contributed to the accident.

Risk management options for application during an aerial application task include reducing the consequence and/or likelihood of adverse events, such as the helicopter striking a powerline. Those options having the potential to decrease the adverse consequences of a wirestrike include:

• the use of helmets, as worn by the pilot
• if appropriate to the helicopter structure, the installation of wire strike protection systems
• the installation of advanced safety harnesses, such as that in the accident helicopter
• appropriate flight following and search and rescue procedures, as in this case due to the ongoing interaction between the pilot and loader/driver at the replenishment point.

However, as was the case in this accident, the consequence of a helicopter striking a powerline can generally be expected to be severe to catastrophic. As a result, a large investment is generally made by involved parties in order to decrease the likelihood, and therefore risk of a wirestrike. In this instance, that included adherence by the pilot, company operations manager and property manager to those regulatory and company requirements affecting the operation, and application by that group of the relevant content of the Aerial Application Pilot’s Manual to the task.

However, by electing to return to the replenishment point via a direct track, and not via the safe transit zone, the pilot unwittingly altered the context or parameters on which his pre-spray application risk assessment was based. The result was that none of the risk treatments applied to the spray operation and safe transit route were applied to the pilot’s approach to the replenishment point.

In that case, the pilot’s return transit was via an unsurveyed route. The dull surface of the powerlines, the nature of the vegetation and topographical background, the location of the supporting poles and long span of the powerlines would have reduced the likelihood that the pilot might have visually acquired the powerlines in sufficient time to have allowed him to avoid striking them.

The investigation was unable to quantify the effect that the installation of high visibility devices on the powerlines, which were not mandated in this case, might have had in preventing this accident. Similarly, the investigation was unable to assess the potential risk reduction capability of the possible installation in the helicopter of an advanced technology wire detection system.

While the reason for the pilot returning to the replenishment point via a direct track could not be determined, the investigation considered it most likely that the pilot had sought to optimise the efficiency of his operation. For example, if the pilot had exhausted the chemical supply at a position where he could observe the location of the replenishment point on the onboard global positioning system equipment, it might have been natural to ‘pop up’ the hill and fly direct to the truck, rather than conduct a long right turn and fly back through the valley. Had the pilot followed the pre-planned safe transit route, or included the direct route from the treatment area to the replenishment point in his risk assessment, he may not have collided with the powerlines.

CONCLUSIONS

Findings

1. The pilot was licensed and qualified for the flight in accordance with existing regulations.
2. The maintenance records indicated that the helicopter was equipped and maintained in accordance with existing regulations and approved procedures.
3. There was no evidence that engine or airframe failure or system malfunction influenced the circumstances of the occurrence.
4. The helicopter’s weight and balance were within prescribed limits.
5. There was no evidence that meteorological conditions influenced the circumstances of the occurrence.
6. The location of the powerlines was outside the safe transit zone between the designated spray area and the replenishment point
7. The powerlines were not marked, nor were they required by the relevant Australian Standards to be marked, with high visibility devices.
8. The position of the powerlines was not depicted on the relevant aeronautical or topographical charts.
9. The company Operations Manual and Aerial Application Pilot’s Manual included advice on the pre-flight planning, briefings and survey necessary before commencing aerial agricultural operations.

Sequence of events¹

On 19 July 2004, the owner-operator pilot of Bell Helicopter Company 47G-3B-1 Soloy² helicopter, registered VH-RTK, was contracted to spray herbicide on a property near Wodonga, Victoria.

Early in the day, the pilot and the company operations manager met with the property owner to discuss the proposed work. The discussion included the identification of known powerlines, other hazards, and sensitive areas likely to affect the operation. Shortly before the discussion, the operations manager showed the pilot the known primary powerlines in the general area where the spraying and loading was to be conducted.

Later that morning, the loader/driver of the chemical and re-fuelling truck arrived at the proposed replenishment point. The pilot positioned the helicopter at the replenishment point at about midday and commenced the final preparations for the day's spraying operations. The preparations included an aerial survey, in the company of the property manager, of the areas to be sprayed. During that flight, the property manager indicated to the pilot the areas that were free from powerlines, including a valley that he considered a safe transit zone between the replenishment truck and the proposed spray area.

The pilot commenced spraying operations on the occurrence property at about 1458 Eastern Summer Time. Shortly before 1512, the property manager heard the helicopter returning to the replenishment point from an unexpected direction and not from the direction of the identified safe transit zone.

The property manager reported that he observed some 'glitter' emanate from the helicopter, before it then descended and disappeared into heavily timbered terrain. No radio broadcasts were made by the pilot to air traffic services, or to the loader/driver to signify that an emergency had occurred.

The helicopter impacted terrain about 860 m to the south-west of the replenishment point, and was destroyed by impact forces. The pilot, who was the sole occupant, was fatally injured. There was no fire.

The loader/driver notified the company operations manager, then commenced to search for the helicopter. The company operations manager notified emergency services, and they arrived at the accident site about 20 minutes later.

A two-cable electric powerline was found severed at a point approximating that at which the property manager reported seeing 'glitter' come from the helicopter. There was no other property damage.

The powerlines struck by the helicopter during the approach to the replenishment point consisted of two parallel three-strand lightweight high-tensile steel cables, each of 2.75 mm diameter. The powerlines had been erected several years previously, and were located on the north-eastern side of a ridgeline, strung across the direct track from the last treatment area to the replenishment point.

The powerlines were not depicted on the relevant aeronautical or topographical charts. They were not connected to the power supply grid and had developed a dull oxidised finish. No high visibility devices were attached to the powerlines, and nor were they required to be.

Personnel information

The pilot was reported to have been fit, well and adequately rested prior to the flight, and he was appropriately qualified and endorsed for the agricultural operation in the helicopter.

Helicopter information

The investigation found that the helicopter was certified, equipped, and maintained in accordance with existing regulations and approved procedures. It was not, nor could it have been, fitted with a wire-strike protection system, nor was there any regulatory requirement for such a system.

No evidence was found of a defect in the helicopter or its systems that may have influenced the circumstances of the occurrence. The weight and balance of the helicopter were within prescribed limits.

The helicopter's perspex windscreen was clean and there was no pre-existing damage, or any obstruction that might have adversely affected visibility from the cockpit. Based on the available evidence, the aircraft was certified, equipped and maintained in accordance with the regulations and approved procedures.

Meteorological information

No evidence was found to suggest that the weather or environmental conditions influenced the circumstances of the occurrence.

Wreckage and impact information

The powerline that was struck lay approximately perpendicular to the helicopter's flightpath from the spray area to the replenishment point (see Figure 1). The powerline was supported by two poles, 669 m apart, which were located on heavily timbered terrain. The point at which the powerline was severed by the helicopter was estimated to be about 34 m above the tree tops, and 54 m above ground level.

Figure 1: Aerial view of powerline and approximate aircraft track

The wreckage distribution and damage to the surrounding trees was consistent with the helicopter impacting the ground with significant vertical, but low forward speed.

A trail of shattered perspex commenced in a position approximately underneath the point at which the powerlines had been severed, and extended for approximately 180 m to the point where the helicopter impacted the ground. The perspex corresponded to that used in the construction of the helicopter windscreen. No other helicopter components were found along the perspex trail.

The ground impact point and wreckage were located on steeply sloping and heavily wooded terrain that was covered with dense vegetation and large rocks (see Figure 2). The area was covered with small to medium sized trees, some approximately 12 to 20 m tall. The ground impact scars indicated that the helicopter was in a relatively level attitude at the time of impact, before it slid approximately 13 to 15 m down the sloping ground and collided with a tree.

Figure 2: Helicopter wreckage

Impact damage to the helicopter was extensive. The helicopter lay on its deformed left side with the cockpit left side panel, the firewall, and the seat assembly compressed in a rearward/inward direction. Damage to the cockpit right side panel was less severe and the panel had retained its shape. A limited amount of the shattered perspex windscreen remained with the main wreckage.

There was severe disruption to the cockpit, and the instrument pedestal had been partially severed from the floor and was pushed towards the pilot's seat. The engine, fuel, hydraulic, and flight control systems' components were dislodged from their installed position and damaged, but remained in their respective locations. The engine was recovered from the accident site and transported to an overhaul facility for technical examination under Australian Transport Safety Bureau (ATSB) supervision. The examination revealed that the engine was capable of normal operation at the time of the accident.

The main rotor blades and mast, main transmission, and the remainder of the helicopter's major components, systems, and controls were also accounted for at the accident site, and in close proximity to the main helicopter wreckage.

Both fuel tanks and their respective firewalls dislodged from their installed position and were damaged. That included impact puncture damage to the fuel tanks. While fuel leakage was evident on the ground around the main wreckage, approximately 20 litres of fuel was recovered from the helicopter's fuel tanks for sampling purposes. A fuel sample was also taken from the re-fuelling truck. The results of independent laboratory tests of both samples were consistent and indicated that the Jet Fuel was bright, free from water and considered suitable for use.

The remains of the tail boom structure was bent slightly upwards, and the tail boom assembly had detached from the rear of the centre fuselage frame about 2 m back from that frame. Sections of the tail boom and the tail rotor drive shaft had separated at impact, but remained oriented in the correct position relative to each other. That was consistent with the helicopter having negligible yaw at impact. A small section of the tail boom structure was located down slope from the main helicopter wreckage.

All major components and extremities of the installed agricultural spray equipment were accounted for with the main helicopter wreckage.

The left and right cockpit door frames were deformed just above the upper door hinge, corresponding to about 15 to 20 cm above the top of the instrument pedestal. That included a deep, circular indentation to the right door frame and a partial tear and bulging of the left door frame. The indentation in the right door frame was 'married' to a length of the severed power cables, and the damage was found to be consistent with the cables having impacted with that door frame (see Figure 3). There were no similar deformations or markings to the instrument pedestal, cyclic or other helicopter controls, or any other part of the helicopter.

Figure 3: The cut into the perspex bubble frame matched with the conductor

Medical and pathological information

Based on the autopsy, toxicology, and medical reports, there was no evidence to indicate that the pilot's performance was degraded by physiological factors.

Survival aspects

A four-point restraint harness, with inertia reel shoulder straps was fitted to the pilot's seat. Video footage of the helicopter departing on the accident flight revealed that the pilot had fastened the harness. Examination of the harness at the accident site revealed that it remained attached to the firewall and that there was no evidence of failure of the locking mechanism.

The pilot was wearing a helmet at the time of the occurrence. That helmet exhibited significant impact damage, but remained intact. The helmet visor was found in the raised position.

Agricultural spray computer disk

The hard disk unit from the on-board agricultural spraying computer was recovered. The unit was assessed by the manufacturer's distributor and an external independent forensic expert. Those examinations confirmed that the damage to the unit precluded extraction of the data.

Organisation

Company Operations Manual

Section A19.5 of the company Operations Manual (manual) contained the en-route procedures affecting low flying operations by company pilots. In part, those procedures included that:

…due consideration shall be given to the dangers of unseen wires, effects of turbulence and other low flying hazards. Prior to the commencement of low level operations, the pilot in command shall carry out an aerial survey of the area to establish an optimum path and aircraft manoeuvring plan for the safe conduct of the operation.

An overview of those procedures affecting the conduct of aerial agricultural operations by company pilots was included at Section D 6.1 of the manual. That included that:

Normal aircraft operating procedures are detailed in the relevant Pilot Operating Handbooks, the Aeronautical Information Publication, as well as CAO 20.21, and the Agricultural Pilots Manual (AAAA [Aerial Agricultural Association of Australia Limited] Pilot and Operations Manual). And these should be followed unless specified otherwise below.

Section D 6.5 of the manual reinforced the regulated requirement for an agricultural pilot to conduct an aerial inspection of a proposed treatment area prior to commencing aerial agricultural as follows:

It is a CASA requirement that prior to commencing agricultural operations the pilot shall carry out an aerial inspection of the proposed treatment area - where practicable, a ground inspection will also be conducted as per sub-section D 6.4.

The manual continued that the pre-treatment aerial inspection was to include the examination of:

…Wires - their location in relation to the boundaries of the area to be treated, height and position of, and distance between poles and guy wires…

Additional information

Previous wirestrike occurrences

A review of the ATSB's accident and incident database revealed that there were six single-fatality agricultural helicopter accidents in Australia during the period 1 January 1995 to 20 June 2005. Those accidents represented 43% of all fatal helicopter wirestrike accidents during the period.

Regulatory framework

Civil Aviation Regulation (CAR) 206 lists agricultural operations as operations conducted for aerial work purposes. Agricultural operations are defined as:

…the broadcasting of chemicals, seeds, fertilizers and other substances from aircraft for agricultural purposes of pest and disease control.

Civil Aviation Order (CAO) 40.6 defines the requirements of the Agricultural Pilot Rating Grades 1 and 2. Those requirements include completion of a period of ground training and a written exam, before carrying out initial and operational flying training. The operational flying training includes operational planning, such as: ground and aerial inspections of the treatment area; assessment of wires; and route selection to and from the treatment area, including the height to fly during such transit flights. That training is to be followed by a period under supervision, before the newly rated agricultural pilot is able to conduct unrestricted agricultural operations.

Aerial Application Pilots Manual³

Chapter 5 of the Aerial Application Pilots Manual contains information pertaining to the identification of powerlines and other low flying hazards with the potential to affect agricultural operations. Chapter 8 of that manual outlines the information specific to helicopter agricultural operations. Those chapters elaborate on the recommended pre-flight planning and inspections that should be carried out by a pilot prior to the conduct of agricultural operations, including that:

If the pilot is obliged to rely on a farmer briefing regarding wire hazards, a thorough interrogation of the farmer is necessary to establish the position of concealed wires…

They are seldom aware that wires outside the treatment area can be hazardous to aircraft approaching for a spraying run. Farmers may forget that they recently connected a shed to the power, ran a new overhead wire to a pump or accidentally hooked a powerline with a farm implement and pulled it over 'a bit'.

Considerable effort is needed to extract vital information from the client. Remember there are some wires that you may strike unless the farmer makes you aware of their location.

In addition, sometimes the background to the wires - trees, hills etc. - provides a poor contrast, while poles may be concealed by intervening obstacles or by being located so far to the periphery of the pilot's visual field that they are not noticed.

During climb and cruise to and from the treatment area the aircraft should track as directly as possible consistent with avoiding nuisance areas and poor terrain.

Don't come below possible powerline height until you are certain all wires have been safely identified.

The property manager reported that he was aware of the powerlines that were struck by the helicopter, but that he had not passed that knowledge on to the pilot because they were located outside both the area to be treated, and the safe transit route between the replenishment truck and the proposed spray area.

Risk management

Australian/New Zealand Standard AS/NZS 4360:2004 Risk Management (the Standard) defined risk as:

the chance of something happening that will have an impact upon objectives.
NOTE 1: A risk is often specified in terms of an event or circumstance and the consequences that may flow from it.
NOTE 2: Risk is measured in terms of a combination of the consequences of an event … and their likelihood…
NOTE 3: Risk may have a positive or negative impact.

The Standard described risk management as 'the culture, processes and structures that are directed towards realizing potential opportunities, while managing adverse [or negative] effects'. The risk management method described by the Standard includes that initially the basic parameters or context affecting the assessment of risk should be identified. That can include 'defining the extent of the project activity or function in terms of time and location'. Residual risk is that 'risk remaining after implementation of risk treatment'. Options for modifying or treating identified risks with negative outcomes included:

• influencing the likelihood of a risk, in order to reduce the probability of a negative outcome
• changing the consequence(s) of an event to minimise the extent of any losses.

Detection of powerlines

The requirements for the mapping and marking of power cables and their supporting structures are published in Australian Standards AS 3891.1 1991 Part 1: Permanent marking of overhead cables and their supporting structures, and AS 3891.2 - 1992 Part 2: Marking of overhead cables for low level flying. The general requirements of those standards were discussed in ATSB  investigation report 200404286 and include that, in general, there is no requirement for the marking of power cables with a height above terrain, or obstacles of less than 90 m. The power cable that was struck by the helicopter did not require marking in accordance with either standard.

Technical committees are formed by Standards Australia to develop and review relevant standards, and comprise a balance of interested and affected parties that are nominated by generally national organisations. The aim is that the standards should include consideration of the views of large, common interest groups. Organisations that consider they represent a valid, previously unrepresented interest group are able to nominate for consideration for inclusion in a committee.

A number of aviation industry associations and other bodies were involved in the development of the Australian Standards affecting the marking of overhead power cables and their supporting structures. That did not include some of the groups and associations normally associated with a number of agricultural and other low-level operations.

There are currently a number of engineering solutions available, with the potential to assist pilots identify overhead powerlines. While their suitability or cost-effectiveness may not prove acceptable for all helicopter types or operations, those engineering solutions include, but are not limited to:

• laser-based systems that alert a pilot of approaching electrified powerlines, or that scan the environment for wires and other obstacles
• enhanced ground proximity-based warning systems that include relevant software and an onboard powerline database in order to identify approaching powerlines.

¹ Only those investigation areas identified by the headings and subheadings were considered to be relevant to the circumstances of the occurrence.
² The designation 'Soloy' indicates that the helicopter had been modified and fitted with a turboshaft engine.
³ Version 6 prepared by the AAAA. A significant upgrade of the Aerial Application Pilots Manual was sponsored by the Civil Aviation Safety Authority (CASA).

The Aviation Safety Investigation Report 200402669 on the fatal accident involving a Bell 47G-3-B-1 helicopter registered VH-RTK which occurred 12km west of Wodonga, Vic. on 19 July 2004. The aircraft was on an aerial agriculture mission when it collided with a powerline. There have been related safety actions from CASA, the Aerial Agriculture Association, and the upcoming ATSB research report on wirestrike accidents.

SIGNIFICANT FACTORS

1. The ADC did not communicate to the crew of the 737 the requirement to pass behind the 717.
2. The crew of the 737 did not perceive the potential threat presented by the 717 until the crew of that aircraft acted to ensure that their flight path did not place them in a near collision situation.

ANALYSIS

The Hamilton Island Aerodrome Controller (ADC) had a plan to separate the aircraft, but did not clearly communicate the plan to the pilots and consequently it was not executed correctly. Having the 737 pass behind the 717 was going to present some difficulties due to the intended tracks of the aircraft, and required the ADC to advise the crew of the 737 as early as possible of that tracking requirement. While traffic information about the 717 was provided to the crew of the 737, the ADC did not communicate an important aspect; that is, that after reporting seeing the 717 they would have to pass behind it. Had that been the case, the crew of the 737 probably would have been able to turn right in sufficient time to safely pass behind the 717. Alternatively, they may have requested another means of separation as the position of the aircraft may have prevented them from passing behind it.

The use of visual separation, either by controllers or pilots, increases the likelihood of an apparent traffic alert and collision avoidance system (TCAS) alert between aircraft. Apparent alerts result from aircraft being within the TCAS alert parameters while complying with an air traffic control (ATC) clearance. In this occurrence, if the 717 crew had not turned away, it is possible that one or both of the aircraft's systems would have issued a traffic alert (TA) or a resolution advisory (RA) In the case of the latter, the pilots would have had to comply. That would have increased crew workload, particularly for the departing 717 crew, when the aircraft was in a critical stage of flight. As it was, the crew of the 717 had to descend to avoid the 737.

The limitations in using visual separation, highlighted in the Manual of Air Traffic Services, applied similarly to the ADC and both crews. The fact that the aircraft were tracking on almost reciprocal tracks, with little or no divergence when viewed by the ADC or the crews, made it difficult for those involved to obtain adequate cues about the situation. However, the crew of the 717 was able to appreciate the potential for conflict by using available visual and system information.

Despite the regulations stating that in a situation where aircraft are approaching head on a pilot shall alter the aircraft's heading to the right, the actions by the crew of the 717, in turning left, could be seen to be reasonable in the circumstances. A turn to the right may have increased the risk of collision. Similarly, if the crew of the 737 had turned their aircraft to the right in accordance with procedures, as the 717 turned left, the risk of a collision may have increased.

Once the ADC was under the impression that responsibility for separation had been transferred to the crew of the 737 and had issued instructions to the crews to climb/descend, the protection afforded by the vertical separation standard was lost. From that point on, the only defences available to the crews to prevent them being in close proximity were their awareness of the other aircraft and the use of TCAS. As they could clearly see the 717, the crew of the 737 did not perceive that there was a problem. The crew of the 717 were concerned at the developing situation and turned away from the 737. That action ensured that sufficient spacing was maintained between the two aircraft and probably prevented a subsequent TCAS RA.

Had the ADC maintained the 1,000 ft vertical separation standard between the aircraft until they had definitely passed, or else confirmed that one of the crews could maintain separation with the other aircraft, it is likely that the occurrence would have been prevented. The use of vertical or lateral separation standards instead of a visual standard would have also limited the possibility of a TCAS alert.

The occurrence highlighted the importance of using unambiguous radiotelephony phraseology to avoid misunderstandings and the need for pilots and controllers to remain vigilant at all times, especially when the dynamics of a situation require action to be implemented early to ensure that the safety of aircraft is not compromised.

FACTUAL INFORMATION

On 17 July 2004, at about 1619 eastern standard time, a Boeing Company 737-476 (737), registered VH-TJH, was inbound to Hamilton Island from the south-east for a landing on runway 14. The Hamilton Island Aerodrome Controller (ADC) instructed the crew to descend to 4,000 ft above mean sea level (AMSL) due to the pending departure of a Boeing Company 717-200 (717), registered VH-VQB, from runway 14. The crew of the 737 requested and were approved by the ADC to track for a left downwind to runway 14. The ADC instructed the crew of the 717 to maintain 3,000 ft AMSL, to make a right turn to track to Mackay and that they were clear for takeoff. The weather was visual meteorological conditions (VMC) and the crew of the 737 reported to the ADC that they could see the 717. The ADC instructed the crew of the 737 to make a visual approach to left base that was amended to a right base after the crew requested that change. Shortly after intercepting the outbound track at about 2,000 ft, the crew of the 717 received a traffic alert and collision avoidance system (TCAS) traffic advisory (TA) and saw the 737 crossing from left to right on descent. The 717 crew's perception was that the expected track of the aircraft would place them on, or close to a collision course so they turned left and descended to avoid the 737 by passing behind it.

Analysis of air traffic control recorded radar data and aircraft flight data revealed that at 1619:15, after the 717 had turned left, the lateral and vertical distance between the aircraft was 1,112 m and 700 ft (737 above the 717). Both aircraft were fitted with a traffic alert and collision avoidance system (TCAS). The flight data recorder (FDR) in the 717 was only capable of recording TCAS resolution advisory (RA) parameters while the 737 FDR did not record any TCAS parameters. Data from the 717 revealed that there was no TCAS RA. The crew of the 717 changed the aircraft's heading by about 35 degrees and descended to 1,500 ft during the manoeuvre, before returning to their assigned track and climbing to 3,000 ft.

Hamilton Island Air Traffic Control

The ADC was responsible for providing air traffic control services in Class D airspace from ground level to 4,500 ft AMSL. In Class D airspace, air traffic control (ATC) is required to separate aircraft operating under the instrument flight rules (IFR) from other aircraft operating under the IFR or special visual flight rules (VFR). In addition, ATC is required to provide the crews of aircraft operating under the IFR with traffic information about aircraft operating under the VFR. The 737 and 717 were both operating under the IFR.

The circumstances were not related to any national airspace changes as both aircraft were IFR and in airspace being managed by ATC.

Clearances and separation

The pilot of the 737 had been issued with a clearance by the Brisbane Centre controller to track inbound to Hamilton Island via the 143 radial of the Hamilton Island VOR¹. The ADC issued a clearance to the pilot of the 717 to track via the 157 radial of the Hamilton Island VOR to Mackay and then the planned route to Sydney. The Aeronautical Information Publication En Route Supplement, Hamilton Island special procedures, nominated a right circuit for operations to runway 14. The use of runway 14 and the issued clearances would result in the aircraft tracks intersecting at some stage (see Figure 1).

Figure 1: Hamilton Visual Terminal Chart with aircraft tracks and times.

The ADC was aware of this and assigned altitudes to the crews that provided the 1,000 ft vertical separation standard required between two aircraft operating under the IFR. The application of that standard was necessary until an alternative separation standard was in place.

When a pilot of an arriving aircraft has been approved by ATC to make a visual approach, they are required to track in accordance with the assigned track clearance until within 5 NM of the aerodrome. From 5 NM, the pilot can diverge from the inbound track to join the circuit as directed by ATC for an approach to the nominated runway.

Air traffic control (ATC) visual separation standards and procedures

The Manual of Air Traffic Services (MATS) details the standards and procedures to be used by controllers to separate aircraft. Part 4, Section 1, Separation Standards stated that in the provision of separation, controllers shall place greater emphasis on traffic planning and conflict avoidance than on individual conflict resolution being achieved. This is to enable separation assurance to be achieved through planning traffic to ensure separation, executing the plan to achieve separation and monitoring the situation to ensure that the plan and the execution are effective. Section 5, Visual Separation stated that visual separation could be achieved by the use of visual procedures (by controllers) or by assigning visual separation responsibility to a pilot. Controllers are to consider aircraft performance characteristics when applying visual separation.

The application of visual separation by the ADC or either crew would have been an appropriate alternative to the vertical separation standard. When aircraft are operating at or below flight level (FL) 125² and will continue to do so during the application of visual separation by a pilot, the pilot of one aircraft is required to report sighting the other aircraft and has to be instructed by a controller to maintain visual separation with, or to follow, that aircraft. Also, if a pilot has been instructed to maintain separation from, but not to follow an IFR category aircraft, then the controller is required to provide traffic information to the pilot of the IFR category aircraft. That information should contain as much detail as possible including the aircraft type, altitude or flight level, position and intentions or direction of flight. If there is any doubt about a pilot's ability to either keep another aircraft in sight or to maintain separation, a controller shall issue alternative instructions to provide separation.

The ADC's reported expectation was that the aircraft tracks would cross such that the 737 would pass behind the 717. At 1615:15, when the 717 was lining up on the runway, the ADC advised the crew of the 737 that the 717 was due to depart direct to Mackay on climb to 3,000 ft and that they could expect to maintain 4,000 ft until the 717 was sighted [by the 737 crew]. The ADC did not advise either crew of the intention, after the 717 was seen by the crew of the 737, to assign separation responsibility to that crew and have them pass behind the 717 (see Appendix A.

While the 717 was taking off, the crew of the 737 reported approaching 4,000 ft and requested a clearance to track for left downwind. The ADC instructed the crew to track for left downwind. At 1618:26, the crew reported to the ADC that they could see the 717. The ADC issued a clearance for the crew to make a visual approach and to report turning a left base. The crew acknowledged the instruction and requested to track to a right base for the runway. The ADC instructed the crew to track for and to report on right base. At 1619:00, the ADC instructed the crew of the 717 to climb to FL310. A pilot approved to make a visual approach can descend as required to establish an aircraft on base or final to the assigned runway. The ADC's instruction to the crew of the 737 to make a visual approach and the instruction to the crew of the 717 to climb, cancelled the separation assurance provided by the application of the vertical separation standard.

Controllers can separate aircraft by visual observation of aircraft position and projected flight paths. The MATS advised that in providing visual separation, controllers should rely primarily on azimuth and if visual separation by judgement of relative distances or altitude is used, then it should be with such wide margins that there is no possibility of the aircraft being in close proximity. The MATS notes that 'experience has shown that a controller's visual determination of the relative distance of aircraft in close proximity can be in error, even to the extent of an apparent reversal of the positions of the two aircraft.' The convergence of the aircraft's respective tracks meant that the lateral spacing and the distance between them, from the ADC's perspective, was reducing.

At 1619:10, approximately 5 seconds before the crew of the 717 elected to turn to avoid the 737, the ADC queried the crew of the 737, 'just confirm that you will be passing behind the 717'. That query was the first time the ADC had mentioned to them that they were required to pass behind the 717. As they replied that they could pass behind, the pilot rolled the 737 right to a maximum bank angle of 26 degrees and 2 seconds later reversed the roll to the left to a maximum bank angle of 15 degrees in conjunction with a left turn when they saw that the 717 was turning away.

At the time of the occurrence, the ADC was managing the two jets and two other light aircraft operating remote from the Hamilton Island Airport area.

Rules of the air

The Civil Aviation Safety Authority Regulation 161 - Right of Way, states that 'an aircraft that is required by the rules in this Division to keep out of the way of another aircraft shall avoid passing over or under the other, or crossing ahead of it, unless passing well clear' and that 'the pilot in command of an aircraft that has the right of way must maintain its heading and speed, but nothing in the rules in this Division shall relieve the pilot in command of an aircraft from the responsibility of taking such action as will best avert collision'.

Regulation 162 - Rules for Prevention of Collision, states that 'when 2 aircraft are on converging headings at approximately the same height, the aircraft that has the other on its right shall give way' and 'when two aircraft are approaching head-on or approximately so and there is danger of collision, each shall alter its heading to the right'.

717 flight crew

The copilot was the pilot flying (PF) and the pilot in command (PIC) was the pilot not flying (PNF). They reported that they were both aware of the inbound 737 and understood the ADC's application of the vertical separation standard. The crew thought that the 737 was probably inbound from Brisbane. Prior to departure, the crew set both cockpit navigation displays³ to 10 NM range in accordance with company procedures.

After take-off, the crew saw on the 717's TCAS display that the 737 was presented as other traffic at about the 11 o'clock position⁴ at approximately 7 NM At that stage they could not see the 737. Shortly after retracting the aircraft's flaps from the take-off position of 18 degrees, they saw the 737 as proximate traffic, slightly above them, on the TCAS. They then saw the 737. The crew became concerned because:

• the 737 was stationary in their windscreen
• the vertical distance between the aircraft was reducing
• the track of the 737 did not appear to be changing
• the 737 was unlikely to be able to pass behind them.

They were also not sure of what action, if any, the crew of the 737 was taking. The copilot had previously reduced the rate of climb because he thought the rate of closure between the aircraft may cause a TCAS alert. The level of concern was such that the PIC instructed the copilot to turn left immediately, and he complied. During the turn the copilot descended the 717 and during that phase the crew received a slats 'overspeed exceedance warning'. The crew reported that the exceedance was 8 kts.

The time between the crew being issued with a clearance to take off and turning left from their outbound track was about 2 minutes.

737 flight crew

The PIC was the PF and the copilot was the PNF. They reported that they understood the effect of the ADC's altitude limits. They saw the 717 back track along the runway and line up as the visibility was 'quite clear.' The crew were initially unconcerned with the departure of the 717 as it was in plain view.

The copilot requested a clearance to track for left downwind in error and 25 seconds later amended it to a request to track for a right base. As the aircraft turned left to track for right base, the 717 was almost stationary in their windscreen where previously, while they were established on the 142 radial, it had appeared to be moving slowly from right to left. They later reported that they eventually realised the potential for conflict after the 717 turned left as they attempted to comply with the ADC's instruction to pass behind it. They received a TCAS TA after the 717 had turned away.

During the period that the crew was requesting approval to track for downwind, the aircraft's flight data indicated that its airspeed was 250 kts and that it had turned left 10 degrees. At that time, it was 9 NM from the aerodrome.

Radiotelephony procedures

Pilots are required to notify receipt of the current terminal information on first contact with ATC, either when taxiing for departure or when inbound for landing. If that advice is not provided, a controller is required to either confirm receipt of the information by the pilot or else provide the current terminal information. Part of the information normally provided is the local altimeter setting. That setting is required by a pilot to enable an aircraft's altimeter to provide the height above mean sea level. It is also needed to ensure the correct application of vertical separation standards between two aircraft. The crew of the 737 did not report receipt of the Hamilton Island terminal information. The ADC did not confirm with the crew that it had been received, nor did the ADC provide the information.

When a pilot is assigned and required to maintain separation with a sighted aircraft, a controller is required to instruct that pilot to, 'maintain separation with (or pass behind or follow) and include details of the aircraft type or identification and any restrictions'⁵. The ADC did not instruct the crew of the 737 to maintain separation or to pass behind the 717 after they reported that they could see that aircraft.

Pilots are required to read back some clearances and/or instructions issued by a controller. Readback items include any altitude or level assignments. If a required readback is not provided by a pilot, a controller is required to challenge the pilot to read back the necessary item. Following the initial inbound report by the crew of the 737, the ADC instructed the crew to descend to 5,000 ft. The crew did not read back that altitude and the ADC did not challenge the lack of a readback. The crew also did not read back the subsequent clearance to make a visual approach. The ADC did not challenge the lack of that readback.

Studies conducted by the US National Transportation Safety Board found that controllers have a tendency to relax their level of alertness in a low workload environment, which makes them vulnerable to operational errors and omissions. Similarly, pilots have been known to relax attentiveness and vigilance when under ATC control.⁶

Traffic alert and collision avoidance system

The Honeywell Incorporated, TCAS II Pilot's Handbook⁷ describes TCAS as:

...an independent on onboard collision avoidance system. It is designed as a backup to the ATC system and the "see-and-avoid" concept⁸. The [TCAS] system is designed to provide safe separation between aircraft predicted to be on collision trajectories while minimising ATC clearance deviation or excursions.

The TCAS does not replace the ATC system. TCAS II continually calculates and tracks the projected positions of air traffic control radar beacon system transponder equipped aircraft within 20 NM and within altitudes 10,000 ft of the aircraft's altitude. The system then generates Resolution Advisories (RA) and Traffic Advisories (TA) against intruder aircraft with ATC transponders.

The level of traffic information displayed is subject to the limits of TCAS, the aircraft's cockpit display and pilot display selections.

A TA is generated for aircraft which are predicted to be within 20 to 48 seconds of the aircraft's collision area⁹, while an RA is generated for an aircraft that is predicted to be within 15 to 35 seconds of the collision area. The timing for an alert is subject to sensitivity levels and altitude layers set in the system. These layers and levels prevent the system from initiating a descent RA when below 1,100 ft above ground level (AGL). For the altitudes of the aircraft at the time of the occurrence, a TA would be generated 25 seconds before the collision area and an RA would be generated 15 seconds before the collision area.

During an RA event TCAS should provide between 300 ft and 800 ft of vertical spacing between the aircraft involved.

¹ VOR - VHF omnidirectional radio range navigation aid.
² 12,500 ft.
³ The displays include details of navigation aids, TCAS and the route to be flown.
⁴ The numbers on a clock are commonly used by a pilot to refer to the relative position of another aircraft. For example, an aircraft observed abeam to the left would be said to be at 9 o'clock.
⁵ Manual of Air Traffic Services, Annex 6-12 Radiotelephony Phrases.
⁶ Shari Stamford Krouse, PhD Aircraft Safety, 1996, ISBN 0-07-036026-X.
⁷ Honeywell Inc, TCAS II Pilot's Handbook, 1.2 Introduction.
⁸ See the ATSB website for further information on see-and-avoid.
⁹ A volume of three dimensional airspace surrounding a TCAS equipped aircraft that varies in size depending upon the rate of closure of a conflicting aircraft.

On 17 July 2004, at about 1619 eastern standard time, a Boeing Company 737-476 (737), registered VH-TJH, was inbound to Hamilton Island from the south-east for a landing on runway 14. The Hamilton Island Aerodrome Controller (ADC) instructed the crew to descend to 4,000 ft above mean sea level (AMSL) due to the pending departure of a Boeing Company 717-200 (717), registered VH-VQB, from runway 14. The crew of the 737 requested and were approved by the ADC to track for a left downwind to runway 14. The ADC instructed the crew of the 717 to maintain 3,000 ft AMSL, to make a right turn to track to Mackay and that they were clear for takeoff. The weather was visual meteorological conditions (VMC) and the crew of the 737 reported to the ADC that they could see the 717. The ADC instructed the crew of the 737 to make a visual approach to left base that was amended to a right base after the crew requested that change. Shortly after intercepting the outbound track at about 2,000 ft, the crew of the 717 received a traffic alert and collision avoidance system (TCAS) traffic advisory (TA) and saw the 737 crossing from left to right on descent. The 717 crew's perception was that the expected track of the aircraft would place them on, or close to a collision course so they turned left and descended to avoid the 737 by passing behind it.

At about 1921 CST, on 13 July 2004, the left engine of a Fairchild Industries Inc. SA227 DC aircraft, registered VH-KEX, failed during the climb phase after take-off. The aircraft was being operated on a scheduled passenger service from Ceduna to Adelaide, South Australia with 2 crew and 17 passengers.

A subsequent examination of the engine, a Garrett TPE331-12UHR, by the operator, in the presence of Australian Transport Safety Bureau (ATSB) investigators, found that the turbine rotating airseal had failed transversely through the cooling holes around the centre of the cylindrical section. That resulted in the liberation of the seal flange and substantial damage to the downstream turbine module (refer Figures 1 and 2).

Since 1995, the engine manufacturer had received four reported failures of the rotating airseal for the TPE331-12 engine. These failures were attributed to progressive thermal cracking. As a result, the manufacturer introduced several service bulletins related to modifications to improve core engine airflow and the introduction of a Fluorescent Penetrant Inspection (FPI) to the rotating airseal. The requirements of these service bulletins had been complied with on this engine during manufacture and FPI was performed at scheduled intervals as required by the manufacturer.

The SMC did not anticipate the aircraft conflict. The night environment, lack of aircraft taxi lights and the distraction with entering system data may have limited his situational awareness despite the availability of the SMR. The use of a segmented or a conditional clearance to the pilots of either aircraft would have assured that the aircraft would not conflict. Alternatively, the SMC could have requested assistance to either enter the system data, or to monitor the taxiing aircraft visually or by using the SMR.

The situational awareness of all the pilots was reduced by the lack of specific traffic information on the other potentially conflicting aircraft. They did not see the other aircraft's taxi light as they carried out their standard lookout. Even if the other aircraft had been seen, it may have been interpreted as an aircraft holding, or even preparing to give way, because the taxi light was selected off.

Both aircraft were being taxied at speeds that limited either crews' ability to react to a conflict in a known busy area of the airport. In this instance a slower taxiing speed would have been appropriate. However, a limiting factor for both aircraft was the lack of a practical means to monitor or assess groundspeed, especially at night.

The SMC's use of the term 'expedite' and a perceived need to not delay taxiing, to assist both pilots and controllers, may have led the pilots of both aircraft to taxi faster than they might normally do. While high taxi speeds may be acceptable in some locations on the movement area, reduced taxi speeds are warranted in potential areas of conflict near runway exits, multiple crossing taxiways and apron access areas. The use of slower taxi speeds in those areas should assist pilots to more readily maintain separation from other taxiing aircraft while complying with air traffic control clearances.

The requirement for pilots to monitor the company frequency throughout the taxi period is a potential distraction for pilots, when the priority for their attention should be on operational duties. Procedures for non-operational radio usage could be better managed so as to have minimal impact on operational duties. This is particularly relevant in a busy taxiway environment like Sydney Airport. If the Dash 8 copilot had also been busy or distracted, a collision may have occurred.

On 2 July 2004, at about 1805 Eastern Standard Time, a de Havilland DHC-8-315 (Dash 8) was taxiing at Sydney Airport for a night departure on a scheduled fare-paying passenger flight to Williamtown, NSW. The Sydney Tower Surface Movement Controller (SMC) had issued a clearance for the crew of the Dash 8 to taxi from Domestic Terminal 1 to the holding point for runway 16 Left. The clearance was via taxiway Charlie, across runway 25 to Bravo 10.

Figure 1: Sydney aerodrome chart

A Saab 340F (Saab) had landed on runway 16 Right and taxied from that runway via runway 25 then turned left onto taxiway Bravo. The SMC issued a clearance for the crew to taxi via taxiway Golf to the apron. The SMC had instructed the crew of the Saab to expedite. The dictionary definition of expedite means to 'speed up the progress of, or to hasten'. The Saab pilot in command (PIC) interpreted this instruction as to not waste any time and to keep the aircraft moving but at a safe speed.

As both aircraft were about to enter the intersection of taxiways Golf and Charlie (see figure 1), the copilot of the Dash 8, seated on the right of the cockpit, saw the Saab. He called for the PIC to stop. The PIC brought the Dash 8 to an abrupt stop as the Saab taxied through the intersection at a constant speed. Neither of the PICs saw the other aircraft until they had crossed at the intersection. The copilot of the Saab was busy with after-landing duties during the initial taxi period after vacating the runway. As the Saab taxied through the intersection he was occupied with a company frequency transmission on the radio and did not see the Dash 8 at any stage. The SMC was unaware of the situation until queried by the PIC of the Dash 8 regarding which aircraft had right of way.

A replay of the surface movement radar (SMR) of the occurrence showed that the taxi speed of the Saab was 26 knots. At 1800:08, the taxi speed of the Dash 8 was 21 knots. It then decelerated and stopped at 1800:11 with the Saab passing directly in front, moving from right to left. The SMR displayed the distance between the aircraft as 42 m. The impression of both PICs was that the aircraft passed in very close proximity.

Normally, pilots use a technique of scanning left and right to check for other aircraft that may conflict with them at intersections. Generally pilots on the left observe the left side and those on the right observe the right side. The standard procedure for the Saab operator was for pilots to survey the area to both the left and the right of the aircraft before entering or crossing any taxiway.

Neither operator had a policy or general guidance material in relation to aircraft taxi speeds, although the Saab operator did require turns in the aircraft to be 'at a speed below 20 kts'. Consequently, the speed at which aircraft were taxied varied between pilots.

Pilots recalled taxi speed limits from their training as 'fast walking pace' but agreed that this was impractical over the long distances involved at Sydney Airport. Neither the Civil Aviation Regulations (CARs) nor the Aeronautical Information Publication (AIP) nominate any specific aircraft taxi speed limits.

A groundspeed readout (utilising a Global Positioning System (GPS)) is available in the Dash 8, but its use is limited by its position on the lower pedestal in the cockpit. This makes it difficult for pilots to cross refer to it while taxiing and looking out for other taxiing aircraft. Similarly, a groundspeed readout is available in the Saab from either the lower pedestal or from the electronic horizontal situation indicator. In addition, a pilot's ability to estimate speed is more difficult at night due to the lack of visual cues used in judging the relative motion of the aircraft with other objects.

The Saab taxi light was unserviceable. The pilots' recollection of the event indicates that the Dash 8 taxi light was not illuminated. Some pilots turn taxi lights off while holding or passing close to other aircraft, to prevent a dazzling effect. The navigation lights and rotating beacons of both aircraft were operating. Both aircraft were backlit to some degree by the domestic terminal and suburban lighting on the far side of the airport.

The SMC was operating the SMC east and SMC west positions on combine. Immediately prior to the occurrence, the controller was endeavouring to manage the arrival and departure from the apron of four aircraft while updating radar system data. System updating included the assigning of labels to radar returns on the SMR for aircraft taxiing for departure. The controller was aware of the disposition of aircraft but his impression, at the time he issued the clearance to the crew of the Dash 8, was that the Saab would pass through the intersection before the Dash 8 would be near that intersection. Consequently, he did not provide traffic information on the other aircraft to either crew, nor did he assess that there was a need to use a segmented or a conditional clearance to either crew. A segmented clearance enables a crew to taxi and to stop at a possible point of conflict, while a conditional clearance enables a crew to taxi subject to specified requirements. In this situation, the Dash 8 crew could have been instructed to taxi and to hold short of the intersection of taxiways Charlie and Golf; or alternatively, to taxi to the runway, with a requirement to pass behind the Saab on taxiway Golf.

Even though crews receive and acknowledge clearances to taxi on the aerodrome manoeuvring areas, the CARs and AIP put the onus on the PIC to maintain a good lookout and observe other traffic to avoid collision.

Both airlines have a procedure where one or both of the pilots listen (and transmit as required) to the company frequency from taxi to just before take off and, from after landing till after shutdown at the parking bay. The Saab copilot was the only pilot reported to be operating a radio at the time of the occurrence.

Sequence of events

On 19 June 2004, at about 1908 Coordinated Universal Time (UTC), two Australian registered Boeing Company 767-338ER aircraft were involved in a serious incident in the Ujung Pandang Flight Information Region (FIR) on air route B473, approximately 60 NM northwest of waypoint SAMGE (latitude 02o21/10//N, longitude 129 o39/00//E).

Flight number QF 83, registered VH-OGP was northbound and flight number QF 98, registered VH-OGQ was southbound on the same air route. The crew of QF 83 reported at SAMGE at 1901 UTC. The crew stated that Ujung Control cleared the crew to `Descend flight level (FL) 350, cross SADAN at FL350 and report leaving FL360'. At about 1906, while continuing to maintain FL360, the crew observed an aircraft (QF 98) on the reciprocal track. The crew of QF 98 were cruising at their cleared level of FL350. This aircraft passed 1,000 ft below QF 83 at about 1908. The crew of QF 98 asked Ujung Control to confirm their clearance. The controller then replied `Maintain FL360, report at time 1917'. If the crew of QF 83 had descended their aircraft from FL360 to FL350 there would have been an infringement of separation standards and an increased risk of collision, with QF 98.

The Australian Transport Safety Bureau (ATSB) was advised of the incident and commenced an investigation. The crews of both aircraft were interviewed and data from each aircraft's quick access recorder was analysed. A review of that data subsequently revealed that the incident occurred inside Indonesian territory. Accordingly, the Indonesian National Transportation Safety Committee (NTSC) had the responsibility to conduct an investigation in accordance with Annex 13 to the Convention on International Civil Aviation.

On 8 July 2004, the NTSC informed the ATSB that they had commenced an investigation into the incident and the ATSB appointed an accredited representative to that investigation. The NTSC, being the investigation agency of the country in which the incident occurred, will be preparing the report and has control over the public release of any investigation findings.

The ATSB will publish the NTSC report on its website www.atsb.gov.au when released by the NTSC.

Factual Information

On the morning of 18 June 2004, a Saab 340 aircraft, registered VH-KEQ, with a crew of three and 31 passengers, was being operated on a regular public transport flight from Albury to Melbourne, Vic. The pilot in command (PIC) had levelled the aircraft at flight level (FL) 120 (12,000 ft) with indicated air speed (IAS) and half bank selected on the autopilot. The engine anti-ice system and propeller and airframe de-ice systems were activated.

The PIC reported that the outside air temperature was -10 ºC, while the IAS was 145 to 150 kts. As the PIC increased the propeller RPM to aid with ice shedding, the IAS rapidly decreased to 137 kts. The PIC disconnected the autopilot and initiated a descent to 10,000 ft. During the autopilot disconnection, the stick shaker activated for about 1 to 2 seconds. The PIC reported that there were no autopilot miss-trim indications during the event. Ice was still present on the aircraft radome after landing.

The stall warning system fitted to the Saab 340 consists of two independent dual channel stall warning computers, left and right angle of attack sensors, two stick shakers and a stick pusher. The system provides five distinct warnings of an impending stall, commencing with stick shaker and aural clacker, followed by autopilot disengage, a visual warning in the form of stick pusher initialisation lights on the instrument panel, and a stick pusher.

The stall warning computers receive inputs from separate angle of attack sensors that are situated on the forward section of the fuselage, which measure airflow relative to the fuselage. Activation of the wing de-ice system increases the angle of attack signal by 0.4 degrees to increase the stall margin by 1 to 2 kts when the de-ice boots are inflated.

The stall warning computer activates the stick shaker at 12.5 degrees angle of attack and the stick pusher at 19 degrees angle of attack, with zero flap deflection and wing de-ice systems deactivated. Activation of the stick shaker causes the autopilot to disengage. Initiation of this warning for both pilots occurs when either of these sensors reaches the predetermined angle of attack. The stick pusher command requires a stall warning output from both sensors, while one or both sensors is required for stick pusher in the event of stall identification.

Following the incident, the data from the aircraft's flight data recorder was downloaded and analysed by the Australian Transport Safety Bureau (ATSB). The data indicated that from the time the aircraft levelled at FL120, the autopilot was maintaining that flight level by providing nose-up elevator movement and automatically re-trimming. At the same time, the IAS was decreasing and the angle of attack was increasing. About one minute later, the RPM of the propellers began to increase from 1,240 RPM to 1,370 RPM. However, over the same period, torque values decreased from 73 to 65 percent and the IAS continued to decrease.

About 1 minute later, with an IAS of 134 kts, the angle of attack reached the value required for stick shaker activation and the autopilot was disconnected. However, because of the disconnection, the subsequent nose-down elevator movements are considered to have been in response to control inputs from the crew, thus the angle of attack did not reach the value required for operation of the stick pusher (19 degrees).

No airframe buffet was evident in the recorded lateral, longitudinal or vertical acceleration data for this event. Recording limitations (sampling rates and the accelerometer frequency response) mean that light buffeting may have occurred and not been evident in the recorded data.

The stick shaker and subsequent control inputs from the crew were initiated before a loss of control of the aircraft.

Recorded data of the incident.

An investigation by the operator following the incident indicated that the probable reason for the rapid decrease in IAS was the altitude capture mode being used at the time of the incident.

As the aircraft approaches the altitude selected on the autopilot, the autopilot will command a capture profile and will hold the selected altitude. The capture point is variable and it is a function of the vertical speed. For this to occur, the autopilot changes the mode from IAS to ALTS (altitude capture) mode, thus giving no airspeed protection during the transition of modes. Without an increase in engine power the airspeed will decrease if the autopilot continues to increase the angle of attack to maintain the captured flight level.

The flight recorder data was also forwarded to the manufacturer to conduct further analysis to establish the reason for the stall and ascertain if the aircraft 'behaved' according to type design. The manufacturer commented that above 10,000 ft, the rate of climb began to decrease and reaching 11,600 ft, the rate of climb had reduced to almost zero. The IAS was 160 kts with propeller RPM 1,240 and engine torque of 69% and 73% on the left and right engine respectively. Thirty five seconds after reaching 11,600 ft the aircraft began climbing at the same time as the indicated airspeed reduced to 150 kts. The aircraft then levelled off at 11,900 ft before making a final altitude adjustment, reaching 12,000 (FL120) at 145 kts, while propeller RPM remained constant at 1,240, but then began increasing about 60 seconds later. Approximately 30 seconds after reaching FL120, the IAS began reducing, until the aircraft entered a stall 100 seconds later. The autopilot, which had been engaged during the climb, was disengaged at the stall warning activation. The aircraft recovered from the stall and descended to 10,000 ft.

The manufacturer commented that the data, illustrated two indications of a stall. The first indication was the increase in angle of attack with no or very small increase in the corresponding lift coefficient. The second indication of a stall was the hysteresis effect in the lift curve seen during the stall. As the aircraft entered the stall and the angle of attack was reduced, the aircraft was not able to attain the normal lift coefficients until the angle of attack was significantly reduced. The analysis also shows that the hysteresis effect was rather moderate, which indicated that the stall had began to build, but was not fully developed, that is, not all parts of the wing were stalled. It is possible that due to the partial stall, the crew may not have recognized it as a stall, especially if control inputs were made simultaneously.

The manufacturer reported that the stall, which happened approximately two and a half minutes after reaching top of climb at FL120, was probably caused by a combination of significant, or extreme, ice accumulation on the airframe, possibly also in combination with run-back ice accretion on the propeller blades. There is an indication from the analysis of the data, that ice was accumulating on the airframe and possibly also on the propeller blades during the final part of the climb above 10,000 ft. The aircraft encountered an aerodynamic stall at the same time as the stick shaker was activated and the autopilot was disconnected. The indicated airspeed at the time of stick shaker activation was 134 kts. The aircraft sustained a moderate roll disturbance to the left during the stall, which was corrected by the crew with moderate opposite aileron deflection. The manufacturer estimated that when the aircraft encountered the stall, the accumulated ice had a combined effect corresponding to a drag increase of more than 500 drag counts, which is in the same order as the total aerodynamic drag for an aircraft without ice accumulation.

The procedure as prescribed in the Aircraft Flight Manual - ref 3 (AFM), as well as in the Aircraft Operators Manual - ref 3 (AOM), is to operate the de-ice boots at the first sign of ice build up anywhere on the aircraft. It is recommended to use the continuous mode of the de-ice boot operation. The continuous mode automatically starts a de-ice boot cycle each 3 minutes and each cycle takes about 30 seconds. However, ice formation on the airframe might in some conditions be so severe that manual de-ice boot operation will be necessary to avoid large ice build-up on the leading edges.

The manufacturer commented that the findings from their aerodynamic analysis show that there was significant, or even extreme, ice accumulation on the wing leading edges as well as other parts of the airframe. It was not possible to determine from the recorded data, if or when the de-ice boot system was operated. Considering the significant increase in aerodynamic drag during the last minutes before entering the stall, the de-ice boot system was probably not operated manually to further enhance the de-icing capability. Had the de-ice boot system been manually and frequently operated during the final part of the climb and during the short cruise segment before entering the stall, ice accumulation would most likely still have been present, but with a significantly less amount and subsequently with less aerodynamic consequence.

According to the manufacturer, the procedures prescribed in the AFM and AOM stated that the propeller de-ice system should be operated in the NORM mode for temperatures between -5 ºC and -12 ºC and MAX mode at temperatures -13 ºC or colder.

Using MAX or NORM modes at warmer temperatures than specified may result in the ice melting, running backwards and refreezing in the form of ridges behind the propeller boots, instead of being shed off the blades. The so-called run back ice will cause a drastic reduction in propeller thrust, up to about 30%.

Above -5 ºC the centrifugal self-shedding capacity is usually enough to avoid ice build-up. Should ice build-up be severe, it is recommended to increase propeller RPM to improve the self-shedding capacity.

From the recorded data, the manufacturer concluded that the aircraft reached an outside air temperature of -5 ºC when climbing through 9,500 ft. Considering the significant loss in rate-of-climb above approximately 10,000 ft, there is a possibility that part of this can be attributed to loss of thrust. The recorded data also reveals that the propeller RPM was low for a climb in icing conditions, approximately 1,240 RPM, which might have reduced the centrifugal self-shedding capacity. The fact that the crew, shortly before entering the stall, increased the propellers to maximum RPM, might indicate that the crew suspected ice formations on the propeller blades.

It is likely that the aircraft had more ice accumulation than the crew realised, which resulted in a degradation of aerodynamic performance that led to a decrease of the IAS, and the subsequent stick shaker activation. Additionally, the crew did not increase power to compensate for the decreasing IAS as the autopilot attempted to maintain altitude by trimming the aircraft to increase angle of attack.

The pilot in command (PIC) of a Saab 340 registered VH-KEQ, had levelled the aircraft at 12,000 feet (FL120) with IAS and half bank selected on the autopilot. The aircraft engine, propeller and airframe anti-ice and de-ice was activated.

The PIC reported that the outside air temperature was minus 10 degrees while the indicated airspeed (IAS) was 145-150 knots. As the PIC increased the propeller RPM to aid with ice shedding, the IAS rapidly decreased to 137 knots. The PIC disconnected the autopilot and initiated a descent to 10,000 feet. During the autopilot disconnection, the stick shaker activated for about 1 to 2 seconds. Ice was still present on the aircraft radome after landing.

The recorded data indicates that from the time the autopilot levelled the aircraft at FL120 the autopilot was maintaining the flight level by providing nose-up elevator movement and automatically re-trimming. At the same time, the IAS was decreasing and the angle of attack was increasing.

About one minute later, with an IAS of 134 knots, the angle of attack reached the level required for stick shaker activation.

A company investigation indicated that the probable reason for the rapid decrease in IAS was most likely caused by the altitude capture mode at the time of the incident.

A manufacturer investigation reported that the wing partially stalled, probably due to a combination of significant ice accumulation on the airframe, and run-back ice accretion on the propeller blades.

The company decided that the current Crew Simulator programme, containing elements of unusual attitude recovery and flight in severe icing conditions, would be extended until the new flight proficiency simulator programme had been trialled and approved. During a visit, the manufacturer informed Australian operators about winter operations in icing conditions.